Distinct spatio-temporal Ca²⁺ signaling elicited by integrin α2β1 and glycoprotein VI under flow

Platelet interaction with exposed extracellular matrix (ECM) at sites of vascular injury is a crucial step in hemostasis and thrombosis.¹  Collagens in ECM mediate both platelet adhesion and activation through direct and indirect mechanisms influenced by fluid dynamic conditions.²  Above a threshold shear rate, the initial interaction between circulating platelets and the vessel wall is mediated by the binding of glycoprotein (GP) Ib to von Willebrand factor (VWF) immobilized onto collagen fibrils.³  The GPIb-VWF interaction promotes the initial tethering, but subsequent firm platelet adhesion is also supported by 2 collagen receptors, GPVI and the integrin α2β1, whose individual roles in collagen binding and platelet activation have been extensively studied in recent years.²  In vivo and ex vivo experiments have suggested that GPVI may be the principal receptor responsible for collagen-induced platelet activation.^2,4  The signaling pathway elicited by the engagement of GPVI is strictly dependent on the Fc receptor γ subunit (FcRγ), which contains an immune-receptor tyrosine-based activation motif and forms a noncovalent membrane-expressed complex with GPVI.² 

The contribution of α2β1 to collagen-induced platelet activation and thrombus formation has been more controversial,⁵  but several observations suggest that it may have an important role. Patients with defective α2β1 manifest a mild bleeding tendency,^6,7  and variations in the expression of this receptor correlate with a predisposition to thrombotic events.⁸  In mice, α2β1 deficiency results in impaired platelet adhesion to collagen and delayed thrombus formation,⁹  although this conclusion may be influenced by the type of thrombosis model used¹⁰  and strain-related differences in its expression are associated with variable response to collagen.¹¹  It is through that, like other integrins, α2β1 requires activation resulting from inside-out signaling as well as divalent cations to engage its ligands with high affinity; and although this may be a requisite for subsequent outside-in signaling, it may not be necessary for initial platelet-collagen contact. Thus, even in a low affinity state, α2β1 may mediate platelet adhesion to collagen preceding GPVI-induced activation.¹²  It is also apparent that α2β1 engagement generates tyrosine kinase-based intracellular signals, which underlie platelet spreading¹³  through a pathway sharing many features with that elicited by GPVI.¹²  Of note, native collagen is an insoluble matrix protein, and the preparations used in ex vivo experiments undergo manipulations that may variably influence the interaction with platelet receptors. For example, α2β1 is required for normal platelet adhesion to pepsin-treated acid soluble collagen but not to acid-insoluble fibrils.¹⁴  Thus, the use of different collagen preparations may explain some of the discrepancies found in the literature with respect to the relative functions of the platelet collagen receptors.

Here, we have used acid-soluble type I collagen and collagen type VI tetramers to study α2β1 and GPVI function under flow conditions. The former collagen type was used to highlight the potential functions of α2β1,¹⁴  the latter because collagen type VI, which forms mixed fibrils with the fibrillar collagens type I and III in ECM,¹⁵  is likely to be readily exposed to flowing blood at sites of vascular injury and, thus, of physiopathologic significance.¹⁶  We found that engagement of α2β1 under flow conditions induces the appearance of transient variations in [Ca²⁺]_i, resulting from store release, and is a requisite for subsequent GPVI-mediated Ca²⁺ signals induced by both collagen types. The sequential function of the 2 receptors underlines a potential synergy in thrombus formation responsive to the collagen composition of the vascular lesion and local fluid dynamic conditions.

Venous blood from medication-free consenting volunteers, in accordance with the Declaration of Helsinki, under protocols approved by the Ethics Committee of Centro Di Riferimento Oncologico and the Institutional Review Board of The Scripps Research Institute, was mixed with one-sixth final volume of citric acid/citrate/dextrose, pH 4.5. The procedures to obtain platelet-rich plasma (PRP), load platelets with the fluorescent calcium probe FLUO 3-acetoxymethyl ester-AM (FLUO 3-AM; Molecular Probes; 8 μM), and prepare a washed erythrocyte suspension have been described previously in detail.¹⁷  In selected experiments, platelets were loaded simultaneously with FLUO 3-AM and BAPTA-AM (1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid acetoxymethyl ester; Molecular Probes; 80 μM). PRP containing 2 to 8 × 10⁸ loaded platelets/mL was mixed with washed erythrocytes, to obtain a suspension with hematocrit of 42% to 45%, and apyrase (grade III; 142 ATPase U/mg protein; Sigma-Aldrich) was added at the final concentration of 5 ATPase U/mL. The mixture was centrifuged at 1000g for 15 minutes, the supernatant was discarded, and the cell pellet was suspended in N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid (HEPES)–Tyrode buffer,¹⁷  pH 7.4, containing 2 mM each Ca²⁺ and Mg²⁺ and 1.75 mM probenecid (Sigma-Aldrich) to prevent FLUO 3-AM leakage from platelets.¹⁸  These procedures did not alter platelet function as evidenced by 3 measurements that were comparable with those of unlabeled platelets: agonist-induced aggregation, P-selectin expression, and binding of IAC-1, a monoclonal antibody that recognizes an activation-dependent epitope on the integrin¹⁹  (kindly provided by Dr Gerlinde Van de Walle, Leuven, Belgium).

All mice experiments were approved by the Institutional Animal Care and Use Committees of Centro Di Riferimento Oncologico and of The Scripps Research Institute. Blood of GPVI null,²⁰  α₂ null,²¹  or wild-type (WT) C57BL/6 mice was collected from the retro-orbital venous plexus using heparin-coated, glass capillary tubes or by cardiac puncture, and mixed with one-eighth volume of citric acid/citrate/dextrose. Mouse PRP was obtained by centrifugation at 100g for 3 minutes at 22°C to 25°C, mixed with an equal volume of HEPES–Tyrode buffer, pH 6.5, containing 5 U/mL apyrase, and centrifuged at 180g for 12 minutes at 22°C to 25°C. The procedure was repeated once, and platelets resuspended at 5 × 10⁸/mL in HEPES–Tyrode buffer, pH 7.4, containing 0.1% bovine serum albumin, 1.75 mM probenecid, 0.01% pluronic acid, were loaded with 16 μM FLUO 3-AM for 40 minutes at 37°C; 1 mL of these labeled platelets was mixed with 5 mL human erythrocyte suspension (50% hematocrit) and 4 mL HEPES–Tyrode buffer, pH 6.5, containing 5 U/mL apyrase. The mixture was centrifuged at 1000g for 10 minutes, the supernatant was discarded, and the cell pellet was suspended in HEPES–Tyrode buffer, pH 7.4, containing 2 mM each Ca²⁺ and Mg²⁺ and 1.75 mM probenecid in a proportion such that the hematocrit was 42% to 45% and immediately used for perfusion experiments.

Acid-soluble type I collagen from human placenta (Sigma-Aldrich) was dissolved in 0.1 M acetic acid at 2.5 mg/mL. Collagen type VI, purified from human placenta as previously reported,¹⁶  was diluted in phosphate-buffered saline (PBS; 20 mM Na₂HPO₄, 20 mM NaH₂PO₄, 2.7 mM KCl, 0.15 M NaCl, pH 7.4) at 200 μg/mL; 100 μL collagen solution was used to coat glass coverslips for 60 minutes at 22°C to 25°C.¹⁴  The GFOGER peptide (GPC[GPP]₅GFOGER[GPP]₅GPC; single-letter amino acid notation, where O = hydroxyproline) was synthesized by SYNPEP Laboratories and cross-linked using N-succinimidyl 3-[2-pyridyldithio]propionate.^22,23  A 50-μg/mL solution of cross-linked GFOGER peptide was incubated on a coverslip overnight at 4°C. After coating, coverslips were washed with PBS and kept in a moist environment until assembled in a modified Hele-Shaw flow chamber.^3,24  In the case of the GFOGER peptide, coated coverslips were treated with 1% fatty acid–free purified bovine serum albumin in PBS for 2 hours at room temperature and washed twice with HEPES–Tyrode buffer before use. The flow chamber was positioned on the stage of an inverted microscope equipped with epifluorescent illumination (Diaphot-TMD; Nikon Instech), an intensified CCD videocamera (C-2400-87; Hamamatsu Photonics), and appropriate filters. The total area of an optical field corresponded to approximately 0.007 mm². Blood cells were aspirated through the chamber with a syringe pump (Harvard Apparatus) at a flow rate calculated to obtain the desired wall shear rate at the inlet. Thrombus volume was measured as reported,^3,25  using blood containing 40 U/mL heparin and 10 μM mepacrine to render platelets fluorescent. Stacks of confocal sections through the height of forming thrombi were obtained at selected time points using a Zeiss LSM 450 inverted confocal microscope. Image analysis for volume calculation was performed with Metamorph (Universal Imaging Co).

When indicated, various reagents were added to the blood cell suspension before perfusion, including PP2, an inhibitor of Src family kinases, or PP3, its noninhibitory analog (Calbiochem-Nova Biochem); PD173952, a different Src family kinase inhibitor (a gift from Pfizer Global Research and Development); wortmannin or LY294002, 2 phosphatidylinositol 3-kinase (PI3-K) inhibitors (Sigma-Aldrich); U73122, a phospholipase C (PLC) inhibitor, or U73343, its inactive analog (Calbiochem-Nova Biochem); MRS 2216, a P2Y₁ inhibitor (kindly provided by Dr K. A. Jacobson, National Institutes of Health, Bethesda, MD)²⁶ ; AR-C69931MX, a P2Y₁₂ inhibitor (Astra Zeneca Pharmaceutical LP)^27,28 ; ethyleneglycoltetracetic acid (Sigma-Aldrich); dibutyryl adenosine 3′,5′-cyclic monophosphate (sodium salt), a cAMP analog, or 8-bromoguanosine 3′,5′-cyclic monophosphate (sodium salt), a cGMP analog, (Sigma-Aldrich); acetyl salicylic acid, an inhibitor of cyclooxygenase-1 and thromboxane A₂ generation, prepared from powdered lysin-acetyl-salicylate (Sanofi-Synthelabo SpA) as a 10-mM stock solution in 0.9% NaCl. Monoclonal antibodies included LJ-CP8, specific for the integrin αIIbβ3 (GP IIb-IIIa complex) and blocking both VWF and fibrinogen binding²⁹ ; Fab 9O12.2, directed against the extracellular portion of human GPVI³⁰ ; R2-7E4, specific for the α2 integrin subunit and blocking α2β1 function¹⁴ ; and TS2/16, specific for the β1 integrin subunit.³¹  Experiments were recorded in real time at the rate of 25 frames/second, which resulted in a time resolution of 0.08 seconds. Selected video sequences were digitized in real time using a Matrox-Digisuite board (Matrox Graphics Inc).

Images were analyzed with a custom-made software (Amplimedical, Casti Imaging Division) that tracks single platelet area, position of the corresponding centroid, and variations of pixel light intensity. Fluorescence intensity variations were converted into [Ca²⁺]_i using the equation:

where K_d is the dissociation constant of the interaction between FLUO 3-AM and Ca²⁺ (864 nM at 37°C),¹⁸  F is the measured fluorescence intensity of a single platelet, F_max is the fluorescence intensity of a single platelet treated with the Ca²⁺ ionophore A23187 (10 μM; Sigma-Aldrich) in the presence of 2 mM CaCl₂, and F_min is the fluorescence intensity of an unstimulated single platelet. The [Ca²⁺]_i of the resting state was calculated in single platelets in which the fluorescence intensity in each of at least 10 consecutive frames was within 15% of the value in the first frame and less than 200 nM. A Ca²⁺ oscillation was recorded (ie, a platelet was activated) when the following conditions were met: a change of [Ca²⁺]_i was more than 3 SD above the resting state value in at least 3 consecutive frames, the magnitude was at least 200 nM, and there was an identifiable peak.³²  A ([Ca²⁺]_i) variation was defined as an α-like peak when the duration was 2.2 ± 0.36 seconds, and a γ-like peak when the duration was greater than 5 seconds. The peak Ca²⁺ concentration was typically greater in the latter than the former, but this parameter was not a discriminator between the 2. Platelets were identified as firmly adherent when present in the first through last frames analyzed (t_x-t₀), typically a time interval of 30 seconds, with centroid movements of less than 1 platelet diameter (2 μm).

Platelet membrane α2β1 density was determined by measuring binding of the monoclonal antibody R2-7E4. For comparison, the binding of LJ-CP8 was measured in parallel. A total of 3 × 10⁶ platelets were incubated with the selected monoclonal antibodies at saturating concentration (75 μg/mL) for 30 minutes at room temperature; polyclonal fluorescein isothiocynate–conjugated rabbit anti–mouse F(ab)₂ (Dako Denmark) was added in a 1:30 vol/vol ratio and incubated for an additional 15 minutes. The samples were analyzed by acquiring a total of 20000 events on a FACScan flow cytometer (BD Biosciences) equipped with CellQuest Pro Software. For analysis, a gate was set around the platelet population as defined by forward and side scatter characteristic, and the mean fluorescence intensity was recorded.

Experiments were performed at least 3 times, and data are shown as the mean plus or minus 95% confidence intervals (CIs) or SEM. Statistical significance was evaluated with the 2-tailed Student t test (degrees of freedom ≥ 10) using Sigma Plot Version 9.0 software (Systat Software Inc).

Flowing platelets interacting with immobilized acid-soluble type I collagen exhibited 2 types of [Ca²⁺]_i oscillations that differed with respect to ion concentration and duration (Figure 1A). One type, which occurred before the other, was characterized by rapid increase to 1 to 2 μM [Ca²⁺]_i lasting for a few seconds and recalled the α peaks seen during GPIbα-mediated platelet rolling onto VWF.³²  These [Ca²⁺]_i peaks were thus defined as α-like. The second type reached the same or higher [Ca²⁺]_i levels but lasted for several seconds, recalling the γ peaks seen during αIIbβ3-mediated stable platelet adhesion to VWF.³²  They were thus identified as γ-like peaks. In 3 experiments, we found that 42.3% plus or minus 6.9% (mean ± 95% CI) of the platelets tethering to acid-soluble type I collagen exhibited α-like and 33.8% plus or minus 5.5% exhibited γ-like [Ca²⁺]_i peaks, the latter being approximately 80% of the former. A function-blocking anti-αIIbβ3 monoclonal antibody (LJ-CP8) had no effects on these Ca²⁺ signals (Figure 1B,E-F; Table 1), whereas an anti-α2β1 monoclonal antibody (R2-7E4) fully inhibited platelet tethering and adhesion to collagen (Figure 1G) and completely blocked all [Ca²⁺]_i oscillations (Figure 1C,E). An anti-GPVI monoclonal antibody (9O12.2), on the other hand, inhibited the initial platelet tethering to collagen (Figure 1G); the appearance of α-like [Ca²⁺]_i peaks (Figure 1D-E; Table 1), and their frequency per platelet (Figure 1F) only partially, albeit significantly, while markedly decreasing the number of firmly adherent platelets (Table 1) and completely preventing the appearance of γ-like [Ca²⁺]_i oscillations (Figure 1D-E; Table 1).

Two different PI3-K inhibitors, wortmannin and LY 294002, equally abolished all Ca²⁺ oscillations and caused a significant, albeit partial, reduction of firm platelet adhesion to acid-soluble type I collagen (Table 1). The latter effect was more pronounced with the PLC inhibitor U73122, which also blocked all Ca²⁺ signals (Table 1). The structurally related but inactive analog U73343 was without effect (not shown). Prostaglandin E₁, an inhibitor of platelet activation acting at different levels, had an effect similar to that of the PLC inhibitor (Table 1). Blocking thromboxane A2 synthesis with acetyl salicylic acid resulted in a modest, albeit significant, reduction in the percentage of platelets exhibiting α-like peaks, but without consequence on the number of platelets exhibiting γ-like peaks or establishing firm adhesion (Table 1). Of note, concurrent blockage of the 2 platelet adenosine diphosphate (ADP) receptors, P2Y₁ with MRS 2216 and P2Y₁₂ with AR-C69931, had no influence on [Ca²⁺]_i oscillations or platelet adhesion (Table 1).

Inhibition of Src family kinases with PP2 resulted in the complete dose-dependent inhibition of both α-like and γ-like [Ca²⁺]_i oscillations in platelets interacting with acid-soluble type I collagen under flow; however, γ-like peaks were abolished at a concentration of inhibitor that only reduced the frequency of α-like peaks by 50% (Figure 2A). A comparable, complete abrogation of α-like peaks required a 5-fold higher concentration of inhibitor (Figure 2A). A high concentration of PP2 (50 μM) also significantly decreased the percentage of firmly adherent platelets (Table 1). PD173952, a structurally unrelated Src kinase inhibitor, at the concentration of 25 μM, demonstrated an inhibitory activity similar to that of PP2 (not shown), whereas PP3, a noninhibitory PP2 analog, was without effect (Table 1). The cGMP stable analog, 8Br-cGMP, also inhibited dose-dependently both types of [Ca²⁺]_i oscillations, and the inhibitor concentration required to abolish α-like peaks was approximately 6-fold higher than that required to abolish γ-like peaks (Figure 2B). At the high concentration of 80 μM, 8Br-cGMP markedly reduced the number of firmly adherent platelets (Table 1). The stable cAMP analog, dibutyryl-cAMP, at the concentration of 100 μM, abolished both α-like and γ-like peaks (not shown).

Platelet adhering to collagen in the presence of 5 mM ethyleneglycoltetraacetic acid (EGTA), a membrane-impermeable Ca²⁺ chelator, showed normal α-like but absence of γ-like [Ca²⁺]_i oscillations (Figure 3), indicating that the latter but not the former depend on transmembrane ion flux. Of note, EGTA had no significant effect on the number of firmly adherent platelets (Table 1). In contrast, all [Ca²⁺]_i oscillations were obliterated by the membrane-permeable chelator, BAPTA-AM, whether in the absence or presence of EGTA (Figure 3), and the number of firmly adherent platelets was significantly reduced (Table 1). Thus, α-like [Ca²⁺]_i peaks are a consequence of release from intracellular stores, which is also a requisite for the subsequent appearance of γ-like Ca²⁺ signals.

Platelets perfused over immobilized TS2/16, an anti-β1 integrin subunit antibody, exhibited frequent α-like [Ca²⁺]_i oscillations on adhesion that reached higher levels than seen on collagen but lacked the sustained duration of typical γ-like peaks (Figure 4A). Platelet incubation with the anti-GPVI monoclonal antibody, 9O12.2, before perfusion had no effect on the appearance of the α-like [Ca²⁺]_i peaks, which were completely inhibited by the Src family kinase inhibitor, PP2, in a dose-dependent manner (Figure 4B). Addition of EGTA had no effect on these [Ca²⁺]_i oscillations (not shown). Platelets perfused over the triple-helical peptide, GFOGER, which interacts specifically with α2β1, exhibited α-like [Ca²⁺]_i oscillations similar to those seen on antibody TS2/16 (Figure 4C). These α-like [Ca²⁺]_i peaks were inhibited by treating the platelets with the anti-α2β1 antibody, R2-7E4, or with PP2 (not shown).

The perfusion of WT mouse platelets over acid-soluble type I collagen resulted in adhesion and appearance of the same 2 types of [Ca²⁺]_i oscillations seen in human platelets (Figure 5A,C). In contrast, the platelets of GPVI^−/− mice exhibited only α-like [Ca²⁺]_i peaks with complete absence of γ-like peaks (Figure 5B), whereas those of α2^−/− mice showed no [Ca²⁺]_i oscillations (Figure 5D-E) while contacting the collagen surface only transiently without achieving firm adhesion (Figure 5F). In 6 distinct experiments, we found that 61.7% plus or minus 12% of WT mouse platelets tethering to acid-soluble type I collagen exhibited α-like and 21% plus or minus 7.6% exhibited γ-like [Ca²⁺]_i peaks (Figure 5E). Fewer GPVI^−/− platelets (39.8% ± 6.5%) exhibited α-like [Ca²⁺]_i peaks with absence of γ-like peaks (Figure 5E), and fewer reached firm adhesion (Figure 5F). Thus, the results obtained with GPVI^−/− or α2^−/− mouse platelets were consistent with those of human platelets treated with the anti-GPVI antibody, 9O12.2, or the anti-α2β1 antibody, R2-7E4, respectively. Thrombus formation in GPVI^−/− or α2^−/− blood was completely abolished (Figure 5G).

Human platelets perfused over purified collagen type VI established firm adhesion essentially in the same proportion (74%) as on acid-soluble collagen type I (75% ± 11%; Table 1) and exhibited the same 2 types of [Ca²⁺]_i oscillations (Figure 6A). Treating platelets with an anti-αIIbβ3 antibody significantly decreased the number of firmly adherent platelets while having no effect on [Ca²⁺]_i oscillations (Figure 6B-C). Thus, to avoid any confounding effect caused by variable αIIbβ3 function in different samples, platelets perfused over collagen type VI were always treated with the anti-αIIbβ3 antibody. When, in addition to the latter, platelets were treated with an anti-α2β1 or anti-GPVI antibody, not only was firm adhesion reduced more markedly but the occurrence of α-like [Ca²⁺]_i peaks was significantly reduced (the effect was more pronounced for the anti-α2β1 antibody) and γ-like [Ca²⁺]_i peaks were essentially obliterated (Figure 6B-C). Concurrent platelet treatment with the anti-α2β1 and anti-GPVI antibodies, in addition to the anti-αIIbβ3 antibody, completely abolished both types of [Ca²⁺]_i oscillations (Figure 6C).

The levels of α2β1 surface expression on the platelet membrane have been reported to vary in normal persons by up to 10-fold,³³  suggesting that there might be corresponding variations in the extent of collagen-induced Ca²⁺ signaling and platelet adhesion. To evaluate this possibility, platelets from persons with either low or high α2β1 membrane density (Figure 7A), but comparable levels of αIIbβ3 (not shown), were perfused over acid-soluble type I collagen. In either case, there was a time-dependent platelet accrual on the surface, but the rate at which this occurred was faster and the extent greater for the high α2β1 density platelets (Figure 7B). Of note, the latter exhibited higher α-like as well as higher and longer-lasting γ-like [Ca²⁺]_i peaks than seen with the low α2β1 density platelets (Figure 7C-D). Treatment with the anti-GPVI antibody before perfusion abolished the occurrence of γ-like peaks and essentially eliminated the difference between high or low α2β1 density platelets, both of which showed only α-like [Ca²⁺]_i oscillations on interaction with collagen (Figure 7E-F).

We identified 2 distinct types of [Ca²⁺]_i elevations in platelets that depend on collagen binding by 2 different receptors, α2β1 and GPVI, under flow conditions. Our results, concordant in antibody-treated human platelets or mouse platelets with receptor ablation, link short-lasting α-like and long-lasting γ-like Ca²⁺ peaks to α2β1 and GPVI ligation, respectively. Such Ca²⁺ peaks are not independent of one another, in agreement with the concept of a collagen-induced reciprocal signaling by integrin and nonintegrin receptors.³¹  Evidence supporting this conclusion derives from experiments performed with acid-soluble collagen type I and fibrillar collagen type VI. In either case, lack of GPVI function has a significant, albeit partial, effect on the appearance of α-like while obliterating γ-like Ca²⁺ peaks; lack of α2β1 function markedly reduces or abolishes α-like and concurrently obliterates γ-like peaks. Thus, α-like [Ca²⁺]_i elevations that require release from intracellular stores, but not γ-like that depend on transmembrane ion flux, can occur in the absence of GPVI, and γ-like can only appear after α-like peaks. This is compatible with the concept that depletion of intracellular stores activates a pathway leading to the opening of plasma membrane channels for regulated Ca²⁺ entry.³⁴  The coupling of α2β1 ligation to intracellular Ca²⁺ oscillations in the absence of GPVI signaling is confirmed by the onset of α-like but not γ-like Ca²⁺ peaks on platelet interaction with an α2β1-specific peptide. Because α-like Ca²⁺ transients predominantly mediated by α2β1 lead to more sustained Ca²⁺ elevations associated with the up-regulation of platelet adhesion and aggregation,³²  our studies highlight a reason why α2β1 may be necessary for full platelet activation. The importance of such a function may vary with the type of collagen involved.

It is apparent from our findings that initial α-like and subsequent sustained γ-like Ca²⁺ transients are neither sufficient nor a requisite, respectively, for firm platelet adhesion to collagen. Indeed, treating human platelets with EGTA abolishes γ-like peaks without significant consequences on platelet adhesion, and blocking GPVI function only modestly inhibits α-like peaks while markedly decreasing platelet adhesion. Moreover, lack of Ca²⁺ signaling is still compatible with firm platelet adhesion to collagen, albeit only at approximately 50% of normal, as seen after intracellular Ca²⁺ chelation or PI3-K blockade. Thus, sufficient activation-inducing signals can occur in platelets to support adhesion against shear stress (which requires platelet spreading)²⁰  without changes in intracytoplasmic Ca²⁺ levels. Of note, our findings indicate that PI3-K as well PLC activity lie upstream of Ca²⁺ release from cytoplasmic organelles induced by the 2 collagen receptors, as both are required to achieve Ca²⁺ elevations; in contrast, only PLC is absolutely required to establish firm platelet adhesion to collagen, with PI3-K contributing to maximal efficiency. Altogether, our studies indicate that distinct changes in [Ca²⁺]_i, detectable in real time and under flow conditions, are markers that differentiate the signaling and adhesive functions of GPVI and α2β1 despite their proposed convergence on the modulation of a similar set of cytoplasmic proteins¹³  during platelet activation.

Models have been proposed to explain how 2 receptors may synergize in mediating collagen responses.^31,35,36  In a 2-step process, α2β1 is thought to initiate the interaction with the substrate, but outside-in signaling required for platelet activation has been considered so far to involve predominantly the GPVI-FcRγ complex.^4,5  Our evidence supports the existence of a functional synergy between α2β1 and GPVI, but not simply to the effect that GPVI provides primary signals subsequently amplified by α2β1.^31,37  Not only did we find that α2β1 can signal independently of GPVI or αIIbβ3 (GPIIb-IIIa),³⁸  as clearly confirmed by the induction of α-like Ca²⁺ peaks on selective α2β1 ligation by the GFOGER peptide, but also that GPVI functions are facilitated by preceding α-like Ca²⁺ elevations dependent on α2β1 engagement. Such conclusions do not exclude the possibility that α2β1 ligand-binding affinity is further enhanced after GPVI interaction with collagen and/or αIIbβ3 activation. With the resolution afforded by our methodology, the functional activities of α2β1 and GPVI appear to be at least concurrent, a concept supported by the finding that variations in α2β1platelet surface expression influence the extent of GPVI-dependent Ca²⁺ signaling. That our findings were not biased by uncontrolled modification of the integrin before exposure to collagen was indicated by the lack of binding of antibody IAC 1, a selective marker of active α2β1 conformation.³⁶ 

Not unexpectedly, we found that cytoplasmic levels of cGMP and cAMP regulate collagen-induced platelet activation, as the same result was previously observed for the 2 corresponding sequential Ca²⁺ signals elicited by GPIbα interaction with VWF under high flow conditions.^17,32 Src kinase inhibitors also display comparable effects on collagen- and VWF-induced activation¹⁷ ; and given the greater inhibitor concentration required to block α and α-like as opposed to γ and γ-like peaks, it appears that the role of Src family kinases varies in sequential steps of activation. In contrast, the role of PLC in collagen, compared with VWF-induced Ca²⁺ signaling, may differ, as in the latter case PLC inhibition caused only partial decrease of initial α peaks but complete abrogation of subsequent γ peaks,⁴⁰  whereas both collagen-induced α-like and γ-like peaks were abolished. The contributions of PI3-K to platelet activation mediated by collagen and its receptors compared with GPIbα and VWF-A1 also appear to differ because, in the latter case, PI3-K inhibitors had no effect on α peaks but prevented the appearance of sustained γ peaks,^17,39,40  whereas both corresponding collagen-induced Ca²⁺ signals were abolished. Thus, Ca²⁺ release from intracellular stores may follow distinct pathways in platelets and involve a variable interplay of PLC- and PI3-K–dependent mechanisms depending on the initiating adhesive event. This may favor a synergy of VWF bound to collagen fibrils in initiating the activation of platelets tethered at sites of vascular injury under high flow conditions. Of note, marking another difference with the process of activation induced by interaction with VWF-GPIb,¹⁷  blockade of both platelet ADP receptors had no effect on Ca²⁺ signals after interaction with collagen. This signifies that secreted ADP, and possibly thromboxane A2, intervenes at a late stage in collagen-induced aggregation.³⁶  In this context, it is difficult to ascertain at present whether the small, albeit significant, effect of aspirin on initial collagen-induced α-like Ca²⁺ peak intensity is relevant with respect to the process of thrombus formation.

In considering the meaning of experimental results obtained with single collagen preparations extracted from tissues, it should be considered that native collagen is an insoluble protein whose thrombogenic activity is probably influenced by multiple interactions with other matrix components in a complex supramolecular assembly.¹  Moreover, evidence is emerging that the function of collagen and/or collagen receptors on platelets may be subjected to the modifying effects of still unknown gene products.⁴¹  Here, to highlight the role of α2β1, we have used acid-soluble collagen type I, which does not exist as such in vivo but is composed of an assembly of fibrils in a helical configuration that may mimic the properties of the spiraled collagens identified in normal and pathologic tissues.⁴²  To extend the significance of our findings, we have also used collagen type VI, which platelets contact directly when the subendothelial matrix is exposed or, associated with collagen type I and III, when lesions reach deeper layers of the vessel wall. In both instances, we found evidence that α2β1 contributes to adhesion and aggregation, but results might be different in the context of platelet interactions with other collagen types and/or with native collagens in tissues. These considerations notwithstanding, our studies on the signaling role of α2β1 contribute to a plausible mechanistic explanation for the increased lag phase and reduced extent of collagen-induced platelet aggregation resulting from functional deficiencies of the receptor, whether caused by inhibitory monoclonal antibodies¹²  or low levels of the protein.^11,43  By the same token, akin to the phenotype resulting from increased cytoplasmic Ca²⁺ levels in mice overexpressing platelet P2×1,⁴⁴  a more rapid and robust α2β1-mediated Ca²⁺ release from intracellular store may contribute to explaining the thrombotic tendency observed in patients bearing the α2 807T polymorphism linked to expression of the receptor.^8,33 

The authors thank Aldo Gasparollo for performing cytofluorimetric analysis.

This work was supported by the Italian Space Agency–Motor and Cardiorespiratory Control Disturbances Project, a Friuli Venezia Giulia Region grant on Mouse Phenotyping, and the National Institutes of Health (grants HL-31950, HL-42846, and HL-78784).

National Institutes of Health

Contribution: M. Mazzucato designed experiments, analyzed data, and contributed to writing the manuscript; M.R.C. performed experiments, analyzed data, and contributed to writing the manuscript; M.B. performed experiments and analyzed data; M.J.-P. and T.J.K. provided essential reagents; M. Mongiat and P.M. performed experiments; Z.M.R. contributed to designing experiments, provided essential reagents, evaluated results, and wrote the manuscript; and L.D.M. supervised research, designed experiments, evaluated results, and wrote the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Luigi De Marco, Servizio Immunotrasfusionale ed Analisi Cliniche, Centro di Riferimento Oncologico National Cancer Institute, Via Franco Gallini no. 2, 33081 Aviano (PN), Italy; e-mail: ldemarco@cro.it; or Zaverio M. Ruggeri, The Scripps Research Institute, MEM 175, 10550 North Torrey Pines Rd, La Jolla, CA 92037; e-mail: ruggeri@scripps.edu.